By Ann Sizemore Blevins
What can each of us do?
Practical Steps. Sometimes supporting diversity, equity, and inclusion (DEI) feels like a big task that requires the effort of people in positions of power (e.g., journal editors, society presidents, department chairs, etc.). But in actuality, there are many practical steps that each of us can take at any level in the academy. Here is a list that we have collated containing simple things that we (and other individual people) can do to support DEI.
We can nominate a Black faculty member (yourself or a colleague) for inclusion in a growing list highlighting the brilliance of Black scientists, engineers, and mathematicians worldwide.
We can look around our institutions and gauge whether the walls reflect the values of diversity, equity, and inclusion. If not, we can consider raising the topic with our administration. In case it is helpful, here is a generic template letter that we recently used to address non-diverse institutional portraiture and other visual arts.
We can use (and speak out for) gender inclusive language, in meetings, in our hallways, and at conferences.
We can start a diversity journal club or otherwise educate ourselves regarding the current science of DEI, recent advances in the field of gender theory or race theory, current evidence for implicit and explicit bias, and current effects of discrimination in the academy and elsewhere.
We can check who we follow on Twitter and surround ourselves with diverse perspectives.
We can restructure the reference list of our current papers or the readings of our current syllabi to cite equitably.
Citation Bias & How to Mitigate It. For scientists active in research and publishing, a simple way to promote DEI is to ensure that the reference list of your manuscripts is balanced in terms of the gender, race, and ethnicity of the cited authors. People from marginalized groups are broadly undercited, by everyone (you and me), impacting visibility, slowing or stalling career advancement, and precluding an unbiased trajectory in the search for scientific truth.
Gender Bias in Citations. Dr. Jordan Dworkin, then a graduate student at Penn and now an Assistant Professor at Columbia University, recently published a paper in Nature Neuroscience reporting gender imbalance in neuroscience reference lists, in a collaboration with philosopher of social justice Dr. Perry Zurn, statisticians Dr. R. Taki Shinohara and Dr. Kristin Linn, and physicist Dr. Erin G. Teich. Follow-up perspectives have appeared in Neuron (“(In)citing action to realize an equitable future”), Nature Neuroscience (“Acknowledging Female Voices”), and Trends In Cognitive Science (“The Citation Diversity Statement: A Practice of Transparency, A Way of Life”), as well as editorials in Brain (see here), Nature Neuroscience (“Widening the Scope of Diversity”), and Nature Reviews Physics (“Growing Citation Gender Gap”). Our findings are consistent with recent papers from other labs reporting gender bias in citations in the fields of astronomy (see here), political science (see here), international relations (see here), cognitive neuroscience (see here), and communications (see here). Journals are responding by calling for action, encouraging authors to include a citation diversity statement, and building a gender balance index into their review evaluations.
Racial/Ethnic Bias in Citations. Dr. Maxwell Bertolero, then a postdoc at Penn, recently posted a preprint reporting racial and ethnic imbalance in neuroscience reference lists, and evaluating their intersection with gender. The work was the culmination of collaborative efforts with Dr. Kafui Dzirasa, Dr. Damien A. Fair, Dr. Antonia N. Kaczkurkin, Dr. Bianca Jones Marlin, Dr. Daphna Shohamy, Dr. Lucina Q. Uddin, Dr. Perry Zurn, Dr. Pragya Srivastava, Sophia U. David, Claudia López Lloreda, Jennifer Stiso, and Dale Zhou. Our findings complement prior evidence from other labs reporting racial/ethnic bias in citations in other fields.
When faced with evidence of bias and discrimination, it is important to consider ways to mitigate it. Here are a few tools developed for that purpose.
Current codebook that can be used to evaluate your own reference list. If you prefer python scripts, see here.
A Google Chrome extension that provides predicted gender of first and last authors of papers in Google Scholar, PubMed, etc.
Citation Diversity Statement. As described and motivated in this piece, consider including a citation diversity statement at the end of your paper, just after the acknowledgments. Here is an example from here. “Recent work in several fields of science has identified a bias in citation practices such that papers from women and other minority scholars are under-cited relative to the number of such papers in the field (1-5). Here we sought to proactively consider choosing references that reflect the diversity of the field in thought, form of contribution, gender, race, ethnicity, and other factors. First, we obtained the predicted gender of the first and last author of each reference by using databases that store the probability of a first name being carried by a woman (5, 6). By this measure (and excluding self-citations to the first and last authors of our current paper), our references contain A% woman(first)/woman(last), B% man/woman, C% woman/man, D% man/man, and E% unknown categorization. This method is limited in that a) names, pronouns, and social media profiles used to construct the databases may not, in every case, be indicative of gender identity and b) it cannot account for intersex, non-binary, or transgender people. Second, we obtained predicted racial/ethnic category of the first and last author of each reference by databases that store the probability of a first and last name being carried by an author of color [7,8]. By this measure (and excluding self-citations), our references contain F% author of color (first)/author of color(last), G% white author/author of color, H% author of color/white author, and I% white author/white author. This method is limited in that a) names, Census entries, and Wikipedia profiles used to make the predictions may not be indicative of racial/ethnic identity, and b) it cannot account for Indigenous and mixed-race authors, or those who may face differential biases due to the ambiguous racialization or ethnicization of their names. We look forward to future work that could help us to better understand how to support equitable practices in science.”
Understanding the Research. Our study of gender imbalance in neuroscience reference lists considered 61,416 articles published between 1995 and 2018. We obtained the predicted gender of the first and last author of each reference by using databases that store the probability of a first name being carried by a woman or by a man. We then calculated the number of cited papers that fell into each of the four first author & last author categories: man & man (MM), woman & man (WM), man & woman (MW), and woman & woman (WW). To obtain the number that would be expected under this assumption of random draws, we calculated the gender proportions among all papers published prior to the citing paper – thus representing the proportion among the pool of papers that the authors could have cited – and multiplied them by the number of papers cited. We defined over/undercitation as the (observed % - expected %)/expected %. By this measure and excluding self-citations, MM papers were overcited by 11.6%, WM papers were undercited by 10.1%, MW papers were undercited by 12.5%, and WW papers were undercited by 30.2%. The effect holds even if we recalculate the base rates of expected citations accounting for 1) the year of publication, 2) the journal in which it was published, 3) the number of authors, 4) whether the paper was a review article, 5) the seniority of the paper's first and last authors (#papers). The effect is largely driven by the citation practices of MM papers; papers that have a woman in the first and/or last author position cite relatively equitably, over-citing MM papers by only 2.5% and under-citing WW papers by only 4.2%. The effect is increasing (not decreasing) with time, particularly in the reference lists of MM papers. Note that the methods we used in this paper are limited in that a) names, pronouns, and social media profiles used to construct the databases may not, in every case, be indicative of gender identity and b) it cannot account for intersex, non-binary, or transgender people.
Our study of racial/ethnic imbalance in neuroscience reference lists considered 63,677 articles published between 1995 and 2019. We obtained predicted racial/ethnic category of the first and last author of each reference by databases that store the probability of a first and last name being carried by an author of color. We then calculated the number of cited papers that fell into each of the four first author & last author categories: white author & white author (NN), author of color & white author (CN), white author & author of color (NC), and author of color & author of color (CC). We measure the expected proportions in the same way as described above. By this measure and excluding self-citations, NN papers are overcited by 7.9%, NC papers are overcited by 1.3%, CN papers are undercited by 6.3%, and CC papers are undercited by 17.2%. The effect holds even if we recalculate the base rates of expected citations accounting for the same 5 factors listed above, as well as the location of the authors' institution. The effect is largely driven by the citation practices of NN papers; papers that have an author of color in the first and/or last author position cite relatively equitably, over-citing NN papers by only 3.2% and under-citing CC papers by only 3.2%. The effect is increasing (not decreasing) with time, particularly in the reference lists of NN papers. Note that the methods we used in this paper are limited in that a) names, Census entries, and Wikipedia profiles used to make the predictions may not be indicative of racial/ethnic identity, and b) it cannot account for Indigenous and mixed-race authors, or those who may face differential biases due to the ambiguous racialization or ethnicization of their names.
When we combine the two pieces of information (gender & race/ethnicity; see Figure 6 in here), we find that papers that have a white man as first author and a white man as last author are overcited by 24%; papers that have a white woman as first author and a white woman as last author are under-cited by 27%; papers that have Black woman as first author and a Black woman as last author are under-cited by 47%. Clearly, these differences could have a marked impact on metrics used to evaluate tenure, promotion, awards, and grant proposals. Note that both of our studies consider approximately the last 25 years, and find increasing under-citation of minorities even as the field has diversified. What do we learn? Greater representation does not itself appear to produce greater engagement with the ideas of scholars from marginalized communities. We also find that co-authorship networks are becoming increasingly segregated by race/ethnicity over the past 25 years; and we tend to cite people nearby on the co-authorship network. What do we learn? Greater representation appears to be tracking greater segregation. We have much work to do.